MUNC18–1 gene abnormalities are involved in neurodevelopmental disorders through defective cortical architecture during brain development

We followed the Fundamental Guidelines for Proper Conduct of Animal Experiments and Related Activity in Academic Research Institution under the jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology, Japan. All of the protocols for animal handling and treatment were reviewed and approved by the Animal Care and Use Committee of Institute for Developmental Research, Aichi Human Service Center (Approval number, M10).

Mouse (m) Munc18–1 clone (mStxbp1) was kindly provided from Dr. T. Izumi (Gunma Univ., Japan) [21]. pCAG-HA-N-Cadherin was from Dr. T. Kawauchi (Inst. Biomed. Res. Innov., Kobe, japan). mMunc18–2, mMunc18–3, mSyntaxin1A and mSyntaxin1B were amplified by RT-PCR with adult mouse brain RNA pool. These cDNAs were constructed into pCAG-Myc vector (Addgene Inc., Cambridge, MA). The following target sequences were inserted into pSuper-puro RNAi vector (OligoEngine, Seattle, WA); mMunc18–1#1, GGACATTGGCACAAGAATA (1558–1574), mMunc18–1#2, GAGGATGAACACTGGCGAG (942–960) and mStx1A#1, GGACATTGGCACAAGAATA (mSyntaxin1A, 1558–1574). Numbers indicate the positions from translational start sites. It should be noted that we could prepare only one RNAi vector specific for mSyntaxin1A, since Syntaxin1A and 1B are small proteins (288 aa) with high homology (84%). For RNAi-resistant versions of mMunc18–1 and mSyntaxin1A, mMunc18–1R and mSyntaxin1A, respectively, silent mutations were introduced in the target sequences as underlined (ATGGGCACTGGCATAAAAACA in mMunc18–1#1; ATGGGCACTGGCATAAAAACA in mStx1A#1). For the control RNAi experiments, we used pSuper-H1.shLuc designed against luciferase (CGTACGCGGAATACTTCGA) [22]. By the use of pCAG-GFP-mMunc18–1R as a template, site-directed mutagenesis was performed with KOD-Plus Mutagenesis kit (Toyobo, Osaka, Japan) to generate Munc18–1 mutants at PKC- and Cyclin-dependent kinase 5(Cdk5)-phosphorylation sites (Ser306/Ser313 and Thr574, respectively), causative mutants for EIEE (c.539G > A, p.C180Y; c.1217G > A, p.R406H; c.1328 T > G, p.M443R) and NEE (c.1631G > T; p.G544 V). Chimeric cDNA, mSyntaxin1AB, containing N- and C-terminal regions of mSyntaxin1A (aa1–149) and B (aa145–255), respectively, was constructed by PCR. All constructs were verified by DNA sequencing.

Mouse monoclonal anti-Munc18–1, anti-Syntaxin1A, anti-nestin and anti-N-Cadherin antibodies were purchased from Abnova Inc. (H00006812-M01; Taipei, Taiwan), Synaptic Systems (110,111; Gottingen, Germany), R&D Systems (MAB2736; Minneapolis, MN) and BD Biosciences (610,920; San Jose, CA), respectively. Mouse monoclonal anti-Myc (#192–3) and polyclonal rabbit anti-GFP (#598) were from MBL Inc. (Nagoya, Japan). Polyclonal rabbit anti-Myc and anti-Sept11 were prepared as previously described [23, 24]. Rabbit polyclonal anti-RFP, anti-HA (hemagglutinin) and anti-N-Cadherin recognizing the extracellular domain were from Rockland Immunochemicals Inc. (#600–401-379; Gilbertsville, PA) and Santa Cruze Biotech. (sc-805 and sc-7939; Santa Cruze, CA), respectively. Polyclonal anti-Tag-1 and monoclonal anti-GM130 were from R&D systems (AF4439; Minneapolis, MN) and BD Biosciences (610,822; San Jose, CA), respectively.

COS7 cells were cultured essentially as previously described [25] and transfected with Lipofectamine 2000 (Life Technologies Japan, Tokyo) according to the manufacturer’s instruction. Immunofluorescence analyses were done as previously described [26]. Alexa Fluor 488- or 568-labeled IgG (Life Technologies Japan) was used as a secondary antibody. 4′,6-diamidino-2-phenylindole (DAPI) was used for DNA staining. Fluorescent images were obtained using FV-1000 confocal laser microscope (Olympus, Tokyo, Japan).

In utero electroporation was performed with pregnant ICR mice essentially as previously described [27]. Briefly, expression plasmids and/or pSuper-RNAi plasmid were injected with pCAG-GFP or pCAG-RFP (red fluorescent protein) into the lateral ventricles of embryos, followed by electroporation using CUY21 electroporator (NEPA Gene, Chiba, Japan) with 50 ms of 35 V electronic pulse for 5 times with 450 ms intervals. As for quantitative analyses of neuronal migration, distribution of GFP- or RFP-positive cells was quantified by calculation of the number of labeled cells in each region of the brain slices [22, 28].

After in utero electroporation, organotypic coronal slices (250 μm thickness) were prepared with a microtome from the anterior third of the forebrain at indicated time points, placed on insert membranes, mounted in collagen gel as previously described [29, 30]. The dishes were then cultured in a CO₂ incubator chamber (5% CO₂, at 37 °C) fitted onto the confocal laser microscope, and the dorsomedial region of the neocortex was examined. Approximately 8–15 optical Z sections were acquired every 5 to 15 min for 24 h, and about 10 focal planes (50-μm thickness) were merged to visualize the entire shape of the cells.

pCAG-GFP vector was electroporated in utero into embryos with control or shMunc#1 at embryonic day (E)14. Forty h after electroporation, pregnant mice were given an intraperitoneal injection of EdU at 25 mg/kg body weight. One h after injection, brains were fixed with 4% paraformaldehyde and frozen sections were made. GFP and EdU were detected with anti-GFP and Alexa Fluor555 azide (Life Technologies Japan), respectively.

Involvement of MUNC18–1 in the etiology of neurodevelopmental disorders implicates its physiological role in brain development. When Munc18–1 expression during mouse corticogenesis was examined by western blotting, it was detected from E13.5 and gradually increased throughout the developmental process analyzed until postnatal day (P)30 (Fig 1a). The expression profile was correlated with that of the northern blotting [31]. In immunohistochemical analyses, Munc18–1 was detected mainly in the intermediate zone (IZ) where axons are enriched and glia cells including oligodendrocytes are not yet present at E17 and P0 (Fig. 1b, Additional file 1: Figure S1). On the other hand, Munc18–1 was detected moderately in the cortical plate (CP) during corticogenesis while it was barely detected in the progenitor and stem cells in the ventricular zone (VZ)/subventricular zone (SVZ) throughout the development (Fig 1b). Notably, Munc18–1 was distributed uniformly in the cerebral cortex in the adult brain (P30) (Fig 1b). These results were consistent with those of in situ hybridization, where Munc18–1 was expressed in CP neurons [32]. Further analyses revealed that Munc18–1 was distributed in the cytosol of bipolar cells committed to layer II-III pyramidal neurons (Fig 1c), which were still migrating in the lower part of CP at E17 as described previously [33]. This result suggests that Munc18–1 participates in radial migration during corticogenesis.

Since Munc18–1 is likely to be involved in the lamination of cerebral cortex during brain development (Fig 1), we examined the role of Munc18–1 in migration of newly generated cortical neurons. We constructed 2 RNAi vectors, pSuper-mMunc18–1(sh-Munc)#1 and #2, which efficiently knocked down Munc18–1 overexpressed in COS7 cells (Fig 2a). These vectors also silenced endogenous Munc18–1 in primary cultured cortical neurons (Fig 2b). Then, pCAG-RFP was electroporated in utero with pSuper-H1.shLuc (Control), sh-Munc#1 or #2 into progenitor and stem cells in the VZ of E14.5 mice brains. When localization of transfected cells and their progeny was visualized at P2, RFP-positive neurons were positioned normally at the superficial layer (bin 1; layers II ~ III) of CP in the control slice (Fig. 2c–d). In sharp contrast, a considerable portion of cells transfected with the RNAi vectors remained in the lower zone of CP and IZ (Fig. 2c,ii,iii and d). Meanwhile, many Munc18–1-deficient neurons still reached the superficial layer of CP (Fig. 2c, ii, iii and d), perhaps due to incomplete depletion of Munc18–1 in neurons incorporating low amount of the RNAi vector; knockdown effects in each cell may vary according to the cell surface area physically exposed to the ventricular lumen from which RNAi vectors pass into cells. Since cell shape is closely associated with cell migration, we examined the morphology of Munc18–1-deficient neurons and found it indistinguishable from normal cells. The deficient neurons apparently formed a normal leading process and attached to radial glial fibers (Fig 2e). We could not examine long-term effects of Munc18–1-knockdown (sh-Munc#1 and #2) since the deficient neurons disappeared at P7 (data not shown), presumably due to cell death as in the case of hippocampal neurons from Munc18–1-null mice [34]. Consistently, caspase3 activation was detected in Munc18–1-deficient neurons at P3 but not during radial migration (E17.5) (data not shown). Notably, brain magnetic resonance imaging (MRI) revealed cortical atrophy in patients with MUNC18–1 mutations, although it is not clear if this atrophy is attributable to the neuronal cell death [35].

Since Munc18–3 is also involved in corticogenesis [7], we asked if the observed migration defects are indeed ascribed to Munc18–1-silencing. Consequently, neither sh-Munc#1 nor #2 silenced mMunc18–3 as well as mMunc18–2 in COS7 cells, strongly suggesting that the observed migration defects were caused by Munc18–1 knockdown (Fig 2f).

At the end of radial migration, the migratory mode changes to the terminal translocation, a crucial step for the completion of neuronal migration, just beneath the marginal zone (MZ) [36]. When we asked if Munc18–1-knockdown affects the terminal translocation, it was completed under the conditions where Munc18–1 was silenced (Additional file 2: Figure S2). The deficient neurons could enter the outermost region of the CP termed primitive cortical zone, and the tip of the leading process was attached to the MZ. These results indicate that Munc18–1 is not involved in the terminal translocation.

Rescue experiments were performed to rule out off-target effects. To this end, Munc18–1R was prepared that was resistant to sh-Munc#1-mediated silencing in COS7 cells (Fig 3a). When pCAG-GFP and shMunc#1 were electroporated with pCAG-Myc-mMunc18–1R, the positional defects were rescued at P2 (Fig. 3b and c), indicating that the mispositioning observed was indeed caused by Munc18–1-knockdown. On the other hand, effects of expression of epilepsy-causative mutants (mMunc18–1-C180Y, −R406H, −M443R or -G544 V) could not be determined since these mutants were presumed to be degradated in cortical neurons as in the case of Neuro2A, PC12 and COS cells (Fig. 3d) [12, 18], suggestive of pathophysiological significance of MUNC18–1 haploinsufficiency for the abnormal cortical neuron migration.

We carried out time-lapse imaging of Munc18–1-deficient cortical neurons migrating in the IZ and CP. VZ cells were electroporated with pCAG-GFP together with control RNAi vector or sh-Munc#1 at E14.5. At the beginning of imaging (E16.5), Munc18–1-deficient neurons displayed multipolar shape similar to the control cells and some cells were transforming into bipolar neurons (Fig. 4a), whereas abnormal phenotypes of the deficient cells came to be observed when time-lapse imaging was continued.

In the control experiment, GFP-positive neurons normally transformed from multipolar to bipolar in the upper IZ and moved into the CP. Then, typical radial migration was visualized toward the pial surface (Fig. 4b upper panels, C left panel and Additional file 3: Video 1). On the contrary, Munc18–1-deficient cells frequently remained stranded in the upper IZ - lower CP during the imaging time period (~24 h) and the migration was significantly prevented during the analyses (Fig. 4b, lower panels, c right panel and Additional file 4: Video 2). Although the multipolar-bipolar shape transition was normal, the deficient cells could not pass the IZ-CP boundary efficiently.

On the other hand, considerable population of Munc18–1-deficient cells crossed the IZ-CP border. They, however, frequently showed an abnormal migration phenotype in the CP. Although the deficient cells were morphologically indistinguishable from control cells and maintained normal bipolar polarity from the beginning of the imaging (E17.5) (Fig. 4d and e), they displayed a migration delay with a reduced speed after entering the CP (Fig. 4e, lower panels, f and Additional files 5 and 6: Videos 3 and 4).

As for the aberrant migration phenotype, it was notably less characteristic than knockdown phenotypes of other genes. For example, SIL1 (a chaperone-related protein)-deficient neurons exhibited abnormal bipolar-multipolar shape turnover during radial migration, while Rbfox1(an RNA-splicing factor)-deficient neurons showed a stepwise migration profile [22, 37]. Meanwhile, Nr1d1 (a circadian clock protein)-knockdown neurons displayed characteristic phenotypes such as jump in a tangential direction and reverse-directed migration [38].

Collectively, Munc18–1 was supposed to be involved in two steps of excitatory neuron migration during corticogenesis; crossing of the IZ-CP boundary and radial migration in the CP. While migration defects may occur at the IZ-CP boundary when the RNAi effect is strong, delayed locomotion in the CP may occur when the RNAi effect is relatively weak.

Since prolonged cell cycle results in a migration delay of newly generated neurons [39], we looked into the effect of Munc18–1-silencing on the cell cycle of progenitor and stem cells in the VZ/SVZ. When S-phase cells were labeled with EdU to detect DNA synthesis, Munc18–1-deficient cells entered S-phase to a similar extent as the control cells (Additional file 7: Figure S3). These results indicate that the cell cycle G1-progression rate was not statistically different between the control and the deficient cells, and that Munc18–1-silencing did not affect cell proliferation in the VZ/SVZ. Also, positioning of EdU/GFP-double positive cells in the VZ/SVZ was not affected by the knockdown (Additional file 7: Figure S3a). Taken together with the data that Munc18–1 was barely detected in the VZ/SVZ (Fig 1b), we conclude that the neuronal positioning defects by Munc18–1-knockdown were most likely to be caused by aberrant migration.

Cdk5 is known to play a central role in neuronal migration during corticogenesis [40] and phosphorylate Munc18–1 at Thr574 to regulate neuronal secretion through the modulation of interaction with Syntaxin 1A [41, 42]. Meanwhile, PKC-mediated phosphorylation of Munc18–1 at Ser306 and Ser313 has been reported to be crucial for its localization in nerve terminals [43], interaction with Syntaxin1A [44] and neurotransmitter release [45]. Analyses with chemical inhibitors suggest possible involvement of PKC in radial migration [46]. We thus analyzed the role(s) of PKC as well as Cdk5 in Munc18–1-mediated cortical neuron migration by examining whether Munc18–1 mutants at the phosphorylation sites by these kinases rescue the knockdown phenotype. As for Cdk5, both Munc18–1-T574D and -T574A mimicking phosphorylated and unphosphorylated states, respectively, rescued the migration defects (Additional file 8: Figure S4a). In contrast, no mutations at the PKC sites mimicking phosphorylated and unphosphorylated states could rescue the abnormal migration phenotype (Additional file 8: Figure S4b). Expression of each mutant was confirmed immunohistochemically (Additional file 8: Figure S4c).

Since Munc18–1 is a regulator for Syntaxin1 in the neurotransmitter release pathway, Syntaxin1 is also possible to regulate neuronal migration downstream of Munc18–1. Interestingly, we found that Syntaxin1A but not 1B rescued the migration defects by Munc18–1-knockdown (Fig. 5a and b), although the rescue effect was less than that by Munc18–1R (Fig. 3b and c). Coordinated function of Munc18–1 with Syntaxin1A was thus suggested to be essential for neuronal migration during corticogenesis. Notably, interaction of Munc18–1 with N-terminal region of Syntaxin1A (aa1–149) might be critical since Syntaxin1AB chimera rescued the migration defects (Fig. 5a–b). Expression of Myc-Syntaxin1A, 1B and AB chimera was confirmed (Fig 5c).

Since Munc18–1-Syntaxin1A interaction was found to be crucial for radial migration during corticogenesis, we examined the role of Syntaxin1A per se in the migration. Syntaxin1A was expressed in the CP, IZ and VZ/SVZ as in the case of Munc18–1 during corticogenesis (Fig. 6a). It was diffusely distributed in the cytoplasm of migrating neurons (Fig. 6b). For the functional analyses, we prepared pSuper-mStx1A(sh-Stx1A), which knocked down exogenous Syntaxin1A but not 1B in COS7 cells (Fig. 6c). Endogenous Syntaxin1A was also knocked, albeit partially, in primary cultured mouse cortical neurons (Fig. 6d). We also made mSyntaxin1A-R resistant to sh-Stx1A-mediated silencing (Fig. 6c). Then, pCAG-RFP was electroporated in utero with the control RNAi vector or sh-Stx1A into the VZ cells of E14.5 mice brains. Although control neurons migrated to the surface of CP at P0, Syntaxin1A-deficient neurons were accumulated in the IZ (Fig. 6e and f). Notably, Syntaxin1A-R rescued the migration defects partially (Fig. 6e and f). Taken together, although the knockdown effects by sh-Stx1A was incomplete, we assume that Syntaxin1A is essential for neuronal migration perhaps downstream of Munc18–1.

Munc18–1 has been shown to be critical for plasma membrane localization of Syntaxin1 in neuroendocrine PC12 cells [47]. To test if Munc18–1 also regulates subcellular distribution of Syntaxin1A in migrating neurons during corticogenesis, we weakly expressed GFP-Syntaxin1A because it is difficult to distinguish the staining signal of endogenous Syntaxin1A in electroporated neurons from that of surrounding cells. pCAG-RFP was electroporated with pCAG-GFP-Syntaxin1A together with the control vector or sh-Munc#1 into E14.5 mouse brain. When cortical neurons were stained for GFP-tag at E18.0, GFP-Syntaxin1A was distributed throughout the cytoplasm in control migrating neurons, whereas it was aberrantly accumulated around Golgi in Munc18–1-deficient neurons (Fig. 7a and b). The ratio of Syntaxin1A signal in perinuclear regions to that of other cytoplasmic regions was increased in the deficient neurons, compared with the control (Fig. 7c). These results indicated that trafficking of Syntaxin1A from Golgi to the plasma membrane was prevented in Munc18–1-deficient migrating neurons. We assume that Munc18–1 plays an important role in the post-Golgi trafficking of vesicles containing Syntaxin1A, which should be essential for radial migration in the CP.

Neuronal cadherin (N-Cadherin) is one of the important adhesion molecules for the attachment of migrating neurons to radial glial fibers. Proper localization and cell surface expression of N-Cadherin is required for glial fiber-guided neuronal migration [48, 49]. We thus analyzed the function of Munc18–1 in vesicle transport by focusing on N-Cadherin distribution. Munc18–1-silencing disturbed distribution of HA-N-Cadherin expressed in migrating neurons and caused its abnormal accumulation at Golgi, compared to the diffuse distribution in the cytoplasm in the control cell (Fig. 8a and b). The ratio of HA-N-Cadherin signal in perinuclear regions to that of other cytoplasmic regions also increased in Munc18–1-deficient neurons (Fig. 8c). These phenotypes were very similar to those of Syntaxin1A trafficking in Munc18–1-deficient neurons (Fig. 7), indicating that Munc18–1 may function in the post-Golgi trafficking of Syntaxin1A- and N-Cadherin-containing vesicles.

Since N-Cadherin exerts its function at cell surface, we next analyzed the involvement of Munc18–1 in vesicle fusion at cell surface. To this end, we used primary cultured cortical neurons to detect cell surface N-Cadherin as neurons were too closely packed in cortical slices to dissect. E14 mouse brain was electroporated in utero with pCAG-GFP together with the control RNAi vector or shMunc#1. After isolation at E16, neurons were cultured for 48 h and then stained for cell-surface N-Cadherin without permeabilization, followed by permeabilization for staining total N-Cadherin. Endogenous N-Cadherin was distributed on the surface of control cell bodies, whereas such distribution was suppressed when Munc18–1 was silenced (Fig. 8d and e). Cell surface distribution of N-Cadherin was also decreased in neurites of the deficient neurons as in the case of cell body (Fig. 8f).

Although transport of Syntaxin1A from Golgi to the plasma membrane region was dependent on Munc18–1, Syntaxin1A per se was not required for the post-Golgi vesicle transport when N-Cadherin was used as an indicator (Additional file 9: Figure S5a, b). On the other hand, Syntaxin1A-knockdown suppressed N-Cadherin distribution on the cell surface as in the case of Munc18–1-silencing (Additional file 9: Figure S5c). Syntaxin1A was thus supposed to regulate the vesicle fusion process to ensure proper interaction of migrating neurons with glial fibers during corticogeneis.

Collectively, while Munc18–1 is known to be essential for the vesicle priming/fusion process in the neurotransmitter release, we here clarified that 1) Munc18–1 likely regulates post-Golgi transport of vesicles containing N-Cadherin, a crucial process for radial migration of excitatory neurons during corticogenesis, and 2) subsequent vesicle fusion and distribution of N-Cadherin to migrating cell surface.